Comparison of Vancouver and International League of Associations for rheumatology classification criteria for juvenile psoriatic arthritis




The International League of Associations for Rheumatology (ILAR) criteria constitute the current international diagnostic standard for juvenile psoriatic arthritis (PsA), replacing the less-restrictive Vancouver criteria. The impact of this change on the population diagnosed with juvenile PsA is unknown.


We reviewed the records of patients seen in a pediatric rheumatology clinic with International Classification of Diseases, Ninth Revision diagnosis codes for psoriasis, PsA, or spondylarthritis. Characteristics of children who met the Vancouver and ILAR criteria were compared.


Of 139 children meeting the Vancouver criteria for juvenile PsA, ILAR criteria excluded 80 (58%). Grounds for exclusion were insufficiently definitive rash (44%), a competing diagnosis of enthesitis-related arthritis (23%), family history of psoriasis limited to second-degree relatives (16%), fulfillment of criteria for >1 subtype of juvenile idiopathic arthritis (JIA) (5%), and HLA–B27 in a male with arthritis onset after age 6 (2%). Remaining patients were not homogeneous but could be divided into younger and older subpopulations differing in clinical features as described previously among patients identified under the Vancouver standard. Of excluded patients, 76% were reclassified as having other forms of JIA yet were phenotypically comparable with those retained.


Despite apparently modest changes from previous criteria, ILAR definitions strikingly restrict the diagnosis of PsA in childhood. Similarity between excluded and included patients suggests that these restrictions may not reflect substantive clinical differences. To the extent that excluded patients become reclassified within JIA, current criteria risk compromising other ILAR categories while reducing the number of patients available for the study of juvenile PsA.


Among the childhood arthritides, juvenile psoriatic arthritis (PsA) is especially challenging to diagnose because the typical psoriatic rash often lags behind the onset of arthritis, occasionally by many years (1–4). While early descriptions of this clinical entity were limited to patients who ultimately developed the classic skin eruption (1, 5–7), later work showed that patients with juvenile PsA could be identified even in the absence of rash by characteristics including dactylitis, nail pitting, an atypical but psoriasiform rash, and a family history of psoriasis (2). Subsequent evaluation of these so-called Vancouver criteria found them to be a robust instrument, internally consistent (demographic and clinical similarity between probable and definite juvenile PsA), and capable of distinguishing patients with juvenile PsA from patients with other forms of juvenile arthritis (Table 1) (4, 8, 9).

Table 1. Vancouver and ILAR criteria for juvenile psoriatic arthritis*
 Vancouver (2)ILAR (12)
  • *

    Arthritis must be of unknown etiology, begin before the sixteenth birthday, and persist for at least 6 weeks. Under the Vancouver criteria, definite juvenile psoriatic arthritis is arthritis plus psoriasis or arthritis plus 3 minor inclusion criteria, whereas arthritis plus 2 minor criteria is considered probable juvenile psoriatic arthritis. ILAR = International League of Associations for Rheumatology; FH = family history; AS = ankylosing spondylitis; ERA = enthesitis-related arthritis; IBD = inflammatory bowel disease; RF = rheumatoid factor; JIA = juvenile idiopathic arthritis.

InclusionArthritis plus psoriasis or arthritis plus at least 2 ofArthritis plus psoriasis or arthritis plus at least 2 of
  Dactylitis Dactylitis
  Nail pits Nail pits or onycholysis
  FH in a first- or second-degree relative FH in a first-degree relative
  Psoriasis-like rash 
ExclusionNoneArthritis in an HLA–B27-positive male beginning after the sixth birthday.
  AS, ERA, sacroiliitis with IBD, reactive arthritis, or acute anterior uveitis, or a history of one of these disorders in a first-degree relative.
  The presence of IgM RF on at least 2 occasions at least 3 months apart.
  The presence of systemic JIA in a patient.
  Arthritis fulfilling ≥2 JIA categories.

However, in an effort to bring the international pediatric rheumatology community to one unified nomenclature, the International League of Associations for Rheumatology (ILAR) proposed (10) and then revised (11, 12) a set of diagnostic criteria aimed at classifying all forms of childhood arthritis into 7 homogeneous, mutually exclusive categories of juvenile idiopathic arthritis (JIA). In the most recent iteration of these guidelines, PsA is defined in a manner that is more restrictive than under the Vancouver criteria (Table 1). Principal differences include the elimination of psoriasis-like rash as a diagnostic criterion, the limitation of family history of psoriasis to first-degree relatives, and the partition of many patients with classic characteristics of the spondylarthritides (enthesitis, onset in the older HLA–B27-positive male child, acute anterior uveitis, inflammatory spine or sacroiliac disease) to the distinct diagnostic category of enthesitis-related arthritis (ERA). Furthermore, patients who do not fulfill criteria for any diagnostic category, or fulfill criteria for more than 1 category, are relegated to the diagnosis of undifferentiated arthritis (6–20% in published series [13–19]).

Reclassification in this way is extremely useful if it serves to divide a heterogeneous group of patients into homogenous subgroups. Indeed, we have recently shown that juvenile PsA as classified under the Vancouver criteria encompasses 2 distinguishable clinical phenotypes, differentiated by age at onset, sex ratio, joint involvement, and clinical course (9). Yet it remains undetermined whether the current ILAR criteria for juvenile PsA demarcate a discrete clinical population. To examine this question, we reviewed the records of 139 children with juvenile PsA as defined under the Vancouver criteria and determined their final classification under ILAR. Our goal was to assess whether the more restrictive ILAR standard improved the classification of these patients by identifying a clinically homogenous population with juvenile PsA while segregating phenotypically distinct patients to other categories within JIA.



We reviewed the charts of every patient seen in the rheumatology clinic of Children's Hospital Boston between January 1997 and February 2005 who had a diagnosis code for psoriasis (International Classification of Diseases, Ninth Revision [ICD-9] code 696.1), PsA (ICD-9 code 696.0), or spondylarthritis (ICD-9 code 756.11). The 139 patients who met the Vancouver criteria were previously described, and because probable and definite juvenile PsA patients were similar across a broad range of demographic and clinical parameters, all were used in the current analysis (9). We subsequently reviewed this cohort to determine whether each patient met the ILAR criteria (Table 1). Of note, a child with onycholysis and 1 additional criterion could meet the ILAR criteria but would not meet the Vancouver criteria; however, we encountered no such patients in our chart review, and therefore the children meeting the ILAR criteria in our sample were entirely subsumed within the Vancouver population.

To determine the frequency with which patients meeting Vancouver and/or ILAR criteria might be missed by this search algorithm, we interrogated the same database for patients meeting other diagnosis codes for arthritis (714, 714.2, 714.30, 714.31, 714.32, and 714.33). This search yielded 1,177 patients over the same interval. From this cohort we reviewed 130 charts to identify 52 patients, not included in the main series, with inflammatory arthritis whose age at onset was <16 years and whose disease lasted at least 6 weeks and had no defined etiology. These charts were reviewed for fulfillment of Vancouver and/or ILAR criteria for juvenile PsA. We identified 2 patients who met the Vancouver standard and 0 patients who fulfilled the ILAR juvenile PsA criteria, consistent with an overall miss rate of 4% (95% confidence interval 0–14%).

The following definitions were used to categorize findings from chart review. A patient was considered to have psoriasis if that diagnosis had been given definitively by a dermatologist or other physician; rashes noted by the examining rheumatologist or reported by the family and thought likely (but not conclusively) to represent psoriasis were considered psoriasis-like. A patient was considered to have nail pits if at least 2 were documented at any visit. Polyarticular arthritis denoted the involvement of ≥5 joints cumulatively at any point over the course of observation; oligoarticular arthritis denoted <5 joints. Small peripheral joints were the metacarpophalangeal and interphalangeal joints of the hands and the corresponding joints of the feet. Large peripheral joints included the wrists, elbows, knees, and ankles. Axial joints included the temporomandibular joint (TMJ), shoulders, cervical or lumbar spine, sacroiliac joints, and hips. Because the inclusion of TMJ as an axial joint is controversial, data for axial excluding TMJ are also provided. Dactylitis was defined as digital swelling extending beyond the margins of the joint. All patients with dactylitis were also considered to have small-joint arthritis in the corresponding digit, although specific attribution to proximal interphalangeal joint or distal interphalangeal joint involvement could not always be made. Enthesitis was any tenderness at tendinous, ligamentous, capsular, or fascial insertions into bone as determined by the examining attending pediatric rheumatologist. Remission was defined as absence of clinically evident synovitis or enthesitis, on or off medications, at any visit and lasting through the last observation for that patient; laboratory parameters were not used. Antinuclear antibody (ANA) values were considered positive if above the upper limits of normal for the laboratory in which the test was performed (generally ≥1:40).

Patients were considered to be excluded under the ILAR standard if they met the criteria for juvenile PsA under Vancouver but not under ILAR. In such cases, the grounds for exclusion were tabulated, and were commonly more than 1 per patient. For example, a child with arthritis in the absence of frank psoriasis but with dactylitis, a family history involving a first-degree relative, and a psoriasis-like rash would qualify for ILAR even though the psoriasis-like rash is discounted; in such cases, the devalued criterion would not be tabulated because it does not affect the patient's classification. However, a similar child with dactylitis, a family history involving a second-degree relative, and a psoriasis-like rash would no longer qualify under ILAR, therefore both family history and psoriasis-like rash would be considered grounds for exclusion.

Statistical analysis.

We compared categorical data and proportions using the chi-square test or Fisher's exact test as indicated. Means were compared with the Student's t-test and medians with the Mann-Whitney U test. Analysis of normality was performed with the Kolmogorov-Smirnov test and Lilliefors correction, with a significant P value indicating evidence of non-normality. Confidence intervals were calculated using the adjusted Wald technique. Because this was an exploratory study, we elected to display differences significant at a 2-tailed P value less than 0.05, uncorrected for multiple comparisons. Institutional review board approval was obtained for this study.


A total of 139 children met the Vancouver criteria for juvenile PsA. Of these, 80 (58%) were excluded under the ILAR criteria. We examined the basis for exclusion of these patients and their final diagnosis under the JIA nomenclature (Table 2). Sixty-one children were excluded on the basis of an inadequately diagnostic psoriasis-like rash, 22 because they had a family history of psoriasis limited to a second-degree relative, and 3 were excluded as HLA–B27-positive males with arthritis onset past age 6. Thirty-two patients were excluded because they fulfilled criteria for ERA (see Table 1, second exclusion criterion); these patients generally manifested arthritis plus enthesitis in the absence of a personal or first-degree family history of psoriasis, which would have excluded them from ERA. Seven children were reclassified as having undifferentiated arthritis because they satisfied inclusion and exclusion criteria for more than 1 subtype of JIA: 3 oligoarticular JIA and juvenile PsA and 4 rheumatoid factor (RF)–negative polyarticular JIA and juvenile PsA. All 7 patients had arthritis with both nail pits and dactylitis, but lacked a personal or family history of psoriasis that would have excluded them from categories other than juvenile PsA. Eleven patients were classified as having undifferentiated arthritis because they failed to fulfill inclusion criteria for any subtype for which they did not also meet exclusion criteria: 2 HLA–B27-positive males with onset past age 6 whose psoriasis excluded them from ERA and 9 patients with a first-degree relative with psoriasis but no other ILAR criteria for juvenile PsA. The final classification of the 80 excluded patients was oligoarticular JIA (n = 23 [29%]), RF-negative polyarticular JIA (n = 7 [9%]), ERA (n = 32 [40%]), and undifferentiated arthritis (n = 18 [22%]).

Table 2. Basis for exclusion of Vancouver patients with juvenile psoriatic arthritis under ILAR criteria*
Reason for exclusionNo. excluded (%)Final ILAR category
Oligoarticular JIARF-negative polyarticular JIAERAUndifferentiated
  • *

    Values are the number unless otherwise indicated. See Table 1 for definitions.

Psoriasis-like rash61 (44)217249
Second-degree FH22 (16)92110
HLA–B27-positive, male, >6 years3 (2)0012
Meets criteria for ERA32 (23)00320
Meets criteria for >1 JIA subtype7 (5)0007
Any80 (58)2373218

Patients included under the ILAR definition for juvenile PsA were compared with patients who were excluded despite meeting the Vancouver criteria (Table 3). Forty-nine (74%) of 66 children with definite juvenile PsA, according to the Vancouver criteria, compared with 10 (14%) of 73 children with probable juvenile PsA were included in ILAR, a difference expected in light of the more conservative nature of the ILAR criteria. Because most patients with frank psoriasis were retained under ILAR, there were significantly more included patients with psoriasis than excluded patients with psoriasis (58% versus 1.3%; P < 0.001). Included patients were older than excluded patients (median age 8.3 years versus 7.0 years; P = 0.02) and exhibited a shorter duration to remission (median 9.7 months versus 19 months; P = 0.03). These latter findings were an expected consequence of the enrichment for patients with frank psoriasis, because younger children more rarely exhibit the classic psoriatic rash while also requiring a longer time to reach remission (9, 20, 21). Otherwise, included and excluded patients were similar with respect to sex ratio, joint involvement, frequency of ANA positivity, and inflammatory markers.

Table 3. Comparison of ILAR-included and ILAR-excluded patients*
 Included (n = 59)Excluded (n = 80)PTotal (n = 139)
  • *

    Values are the percentage unless otherwise indicated. IQR = interquartile range; TMJ = temporomandibular joint; ANA = antinuclear antibody; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; see Table 1 for additional definitions.

  • Differences significant at P ≤ 0.05.

  • Among those with >6 months of followup. Included: n = 43; excluded: n = 68.

  • §

    Among those with >1 visit. Included: n = 53; excluded: n = 76.

Female sex615859
Psoriasis581.3< 0.00125
Age at onset, median (IQR) years8.3 (4.1–12)7.0 (1.8–10)0.0207.3 (2.5–11)
Joint involvement    
 Axial involvement152420
 Axial excluding TMJ121815
 Any small joint565857
 Any large joint808080
Oligo onset928889
Oligo at 6 months888184
ANA positive, no./total no. (%)18/46 (39)36/71 (51)54/117 (46)
Uveitis, no./total no. (%)1/24 (4.2)5/52 (9.6)6/76 (7.7)
Inflammatory markers, mean ± SD    
 ESR, mm/hour23 ± 1527 ± 2325 ± 20
 CRP, mg/dl0.37 ± 0.70.72 ± 10.59 ± 1
 Platelets, 103/mm3342 ± 101375 ± 101361 ± 102
Outcome at last visit§    
 Clinical remission496357
 Duration to remission, median (IQR) months9.7 (2.3–24)19 (5.6–49)0.03015 (4.1–42)

We have previously shown that patients with juvenile PsA diagnosed according to the Vancouver criteria constitute 2 subpopulations, with clear differences based upon age at onset (9). To determine whether the exclusion of nearly 60% of the Vancouver population led to increased homogeneity within juvenile PsA under ILAR, we examined the age-at-onset curve of the ILAR population and compared it with that of the Vancouver group. As shown in Figure 1, the histograms remained similar. In both groups there was a peak among children under 5 years of age, with a less well-defined older population. The distribution of age at onset among ILAR patients failed the Kolmogorov-Smirnov test for normality (P < 0.001, Lilliefors correction), as reported previously for the Vancouver population (9). We compared the clinical features of children whose age at onset was <5 years with those whose age at onset was ≥5 years (Table 4). We found that older and younger ILAR children with juvenile PsA differed over many of the same clinical variables as previously identified within the Vancouver group (9). Thus, younger children were more likely to be female (87% versus 52%; P = 0.018), to exhibit dactylitis (60% versus 25%; P = 0.013), and to have elevated inflammatory markers at onset (mean erythrocyte sedimentation rate 31 mm/hour versus 20 mm/hour; P = 0.031). Likewise, older children were significantly more likely to have enthesitis (54% versus 13%; P = 0.005). Differences in the incidence of axial involvement, ANA positivity, and polyarticular onset and course were not statistically significant in this smaller population, although trends remained consistent with findings in the larger Vancouver cohort.

Figure 1.

Distribution of age at onset among patients diagnosed with juvenile psoriatic arthritis under the International League of Associations for Rheumatology (ILAR) criteria (n = 59) compared with the Vancouver population as a whole (n = 139 [9]). Age numbers refer to years of life (for example, 1 = age 0–364 days).

Table 4. Comparison of ILAR-included patients with juvenile psoriatic arthritis under and over age 5 years*
 <5 years (n = 15)≥5 years (n = 44)P
  • *

    Values are the percentage unless otherwise indicated. IQR = interquartile range; TMJ = temporomandibular joint; ANA = antinuclear antibody; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; see Table 1 for additional definitions.

  • Differences significant at P ≤ 0.05.

  • Among those with >6 months of followup.

  • §

    Among those with >1 visit.

Female sex87520.018
Age at onset, median (IQR) years2.3 (1.4–2.6)11 (7.0–13)< 0.001
Joint involvement   
 Axial involvement020
 Axial excluding TMJ016
 Any small joint6752
 Any large joint8777
Oligo onset9391
Oligo at 6 months, no./total no. (%)9/11 (82)29/32 (91)
ANA positive, no./total no. (%)6/13 (46)12/33 (36)
Uveitis, no./total no. (%)1/11 (9.1)0/13 (0)
Inflammatory markers, mean ± SD   
 ESR, mm/hour31 ± 1520 ± 140.031
 CRP, mg/dl0.45 ± 0.60.35 ± 0.7
 Platelets, 103/mm3381 ± 109331 ± 97
Outcome at last visit§   
 Clinical remission, no./total no. (%)6/13 (46)20/40 (50)
 Duration to remission, median (IQR) months9.6 (3.5–36)10 (2.1–27)

To assess whether patients reclassified to other ILAR subgroups were distinct from those retained under juvenile PsA, we examined these groups for clinical and demographic differences beyond those imposed by subgroup definitions. We found that such differences were minimal, although our power was limited by group size (Table 5). The apparent exception was patients reclassified as having polyarticular JIA, who were considerably younger (1.4 years versus 8.3 years), had significantly more elbow, wrist, and small-joint involvement compared with the ILAR group, and were more likely to be ANA positive. However, 6 of 7 patients in this group exhibited dactylitis and the seventh patient had nail pits, findings more typical of juvenile PsA than nonpsoriatic JIA (8, 22, 23). Among the 17 patients with ERA, 4 had experienced occult, asymptomatic iridocyclitis, and none had developed acute anterior uveitis.

Table 5. Comparison of reclassified patients with the ILAR juvenile psoriatic arthritis cohort*
 Psoriatic JIA (n = 59)Oligoarticular JIA (n = 23)Polyarticular JIA (n = 7)ERA (n = 32)Undifferentiated arthritis (n = 18)
  • *

    Values are the percentage unless otherwise indicated. IQR = interquartile range; TMJ = temporomandibular joint; ANA = antinuclear antibody; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; see Table 1 for additional definitions.

  • Differences from the ILAR group significant at P ≤ 0.05.

  • Among those with >6 months of followup.

  • §

    Among those with >1 visit.

Female sex61441005367
Age at onset, median (IQR) years8.3 (4.1–12)6.7 (2.1–8.9)1.4 (1.2–4.7)9.1 (4.4–11)2.5 (1.5–9.8)
Joint involvement     
 Axial involvement1513432528
 Axial excluding TMJ128.7291628
 Any small joint56481005656
 Any large joint80741008452
Oligo onset92100149489
Oligo at 6 months, no./total no. (%)38/43 (88)20/20 (100)0/6 (0)23/26 (88)12/16 (75)
ANA positive, no./total no. (%)18/46 (39)12/21 (57)6/6 (100)11/29 (38)7/15 (47)
Uveitis, no./total no. (%)1/24 (4.2)0/16 (0)1/5 (20)4/17 (24)0/14 (0)
Inflammatory markers, mean ± SD     
 ESR, mm/hour23 ± 1526 ± 2928 ± 2426 ± 1830 ± 26
 CRP, mg/dl0.37 ± 0.70.76 ± 10.71 ± 0.90.61 ± 0.70.90 ± 2
 Platelets, 103/mm3342 ± 101361 ± 111366 ± 121378 ± 103388 ± 90
Outcome at last visit§     
 Clinical remission, no./total no. (%)26/53 (49)14/22 (64)4/6 (67)16/30 (53)14/18 (78)
 Duration to remission, median (IQR) months9.7 (2.3–24)29 (4.3–50)24 (14–48)19 (7.7–62)16 (3.6–48)

Given these indications that the current ILAR standard for juvenile PsA is overly restrictive, we explored criteria modifications that might limit the exclusion of patients from this diagnosis. Because 61 patients were eliminated at least in part on the basis of insufficiently diagnostic skin rashes, we attempted to stratify these rashes according to the degree to which they were likely to be informative of an underlying psoriatic diathesis. However, in this retrospective study, we were unable to identify significant clinical differences based upon rash description (data not shown). Similarly, a family history of psoriasis in second-degree relatives can be difficult to obtain with accuracy and may encompass >10% of children presenting with arthritis (12, 16) (ML Stoll, PA Nigrovic, unpublished observations), so we did not attempt to rescue this Vancouver minor criterion.

However, nail pits and dactylitis are potentially more informative. Although nail pits may result from a number of causes, including trauma and alopecia areata, their presence in a child with arthritis is strongly suggestive of psoriasis (23). Similarly, PsA is the hallmark disease in which dactylitis occurs in children (24–26). We therefore explored the effect of equating nail pits and/or dactylitis with psoriasis for the purpose of juvenile PsA diagnosis. Accepting nail pits as equivalent to psoriasis restored 38 of 80 excluded patients to juvenile PsA, including 18 of 32 patients reclassified as having ERA. Accepting dactylitis in the absence of an alternate cause as equivalent to psoriasis restored 31, including 12 ERA patients, while admitting either nail pits or dactylitis as equivalent to psoriasis restored 58 of 80 excluded patients (26 of 32 ERA). The 58 children who were reincluded, as compared with the 22 who were not, tended to be younger (median age 5.3 years versus 7.8 years; P = 0.180), female (62% versus 46%; P = 0.180), to exhibit less axial involvement (excluding TMJ, 12% versus 32%; P = 0.051), and to have polyarticular disease (17% versus 0%; P = 0.054). They also demonstrated a longer median duration to remission (31 months versus 8 months; P = 0.014). Thus, the restored children belonged predominantly to the younger subgroup identified previously (9), which had been excluded disproportionately by the ILAR criteria because younger children are less likely to exhibit frank psoriasis.

Finally, we investigated the impact of a nail pits/dactylitis criterion on the diagnosis of inflammatory arthritis in patients not currently meeting Vancouver or ILAR criteria for juvenile PsA and therefore not included in this series. Of 50 patients randomly selected from ∼500 children diagnosed with arthritis over the study interval but who failed to meet Vancouver criteria, 2 patients met this new criterion. Both had mixed large- and small-joint disease affecting 4–6 joints in an asymmetric distribution. One patient's father had a postauricular scaling rash but did not carry a formal diagnosis of psoriasis. In both cases, juvenile PsA was the attending pediatric rheumatologist's working diagnosis.


The classification of juvenile arthritis is important for effective communication among physicians as well as to enable multicenter cooperation in the study of these rare diseases (27). Because subgrouping based on shared pathophysiology is not yet possible, classification systems will necessarily depend upon clinical phenotype. Such systems must walk a narrow line. If criteria for a disease category are too loose, then the population identified becomes too inhomogeneous to study profitably. If criteria are too strict, then individuals are excluded unnecessarily and it becomes difficult to assemble patients for retrospective review or prospective research. Improperly excluded patients may also become included in other diagnostic categories within the system, complicating the study of those conditions. Analogous problems face the adult rheumatology community in the establishment of classification criteria for adult PsA (28).

Understanding that there is no gold standard for the classification of juvenile arthritis, we tested the international consensus ILAR standard for juvenile PsA in 3 ways. First, we compared patients with Vancouver juvenile PsA who met the ILAR criteria for juvenile PsA with patients who did not. Second, we examined the internal homogeneity of the population identified as having juvenile PsA under ILAR. Third, we compared patients included under Vancouver but excluded under ILAR with those retained within juvenile PsA. We hypothesized that the ILAR criteria, if successful, would 1) retain most patients diagnosed under Vancouver; 2) identify a clinically and demographically homogeneous population, especially if a substantial proportion of Vancouver-defined patients were eliminated; and 3) exclude only patients with clinical features that distinguish them from patients retained within juvenile PsA, partitioning these patients appropriately into other categories under the JIA nomenclature.

Despite apparently minor changes in classification criteria, we found that the ILAR criteria excluded almost 60% of patients included under the Vancouver standard. Even with these exclusions, ILAR patients with juvenile PsA remained inhomogeneous, with clinically distinct younger and older patients similar to those identified within the Vancouver population (9). Aside from being slightly younger, excluded patients were similar to the included population across a broad range of demographic, clinical, and laboratory parameters (Table 3). To understand whether this observation might miss the legitimate elimination of one or more small but clinically distinct subgroups, we compared excluded patients assigned to each JIA class with the ILAR juvenile PsA pool (Table 5), finding few differences between juvenile PsA and the other JIA subclasses beyond those introduced as a function of the entry criteria for each subset.

Indeed, some characteristics of the excluded patients appeared to strongly favor their continued inclusion within juvenile PsA. For example, all 7 patients reclassified as having polyarticular RF-negative JIA had either dactylitis or nail pits, whereas 4 of 17 patients reclassified as having ERA had experienced occult, asymptomatic iridocyclitis of the type most characteristic of oligoarticular and psoriatic JIA, and none had developed the acute painful anterior uveitis of the classic spondylarthritides (22).

In this population of 139 patients, it therefore appears that the ILAR criteria do not accomplish the goal of segregating patients into clinically distinct, internally homogeneous groups. Furthermore, our data suggest that present ILAR definitions place patients with arthritis indistinguishable from juvenile PsA into other categories of JIA, most prominently ERA. As a consequence, the prevalence of juvenile PsA is probably substantially underreported in recent international series (13, 14, 17, 29, 30). While it remains possible that the Vancouver criteria cast too broad a net, labeling individuals with other JIA subtypes as having juvenile PsA, it seems clear that strict application of the ILAR definitions in real practice leads to the misclassification of a substantial proportion of children with juvenile PsA.

How could the ILAR criteria be amended to better accommodate juvenile PsA? One possibility would be to admit nail pits or dactylitis as equivalent to psoriasis in children with arthritis, since these features strongly suggest juvenile PsA. In our database such a change would have restored almost 75% of patients, while excluding the large majority of patients with nonpsoriatic juvenile arthritis. Pattern of joint involvement may be another important clue to the diagnosis of juvenile PsA, as has been suggested previously (8, 22), but it is beyond the scope of the present work.

The incorporation of psoriasis-like rashes is potentially more complicated. Young children with psoriasis tend to present atypically, and the dermatologic diagnosis is often challenging (21). From our review of rashes reported in the medical records, we suspect that some skin findings were identified and documented by the examining physician only because of a clinical suspicion for juvenile PsA based on factors such as the presence of nail pits, family history, or pattern of joint involvement. We believe that the development of accurate criteria for juvenile PsA, including the changes we have considered above, will require careful prospective data collection in a population of children with both psoriatic and nonpsoriatic arthritis.

Our results highlight the need for caution in the application of the ILAR category of ERA. Published data indicate that there exists a population of children who present with enthesitis as a central feature of their clinical syndrome, and who are distinct from other patients with juvenile arthritis in that they are commonly HLA–B27 positive and at high risk of developing axial disease (31–33). However, enthesitis is a well-recognized feature of PsA (34). If it is meaningful to distinguish types of childhood arthritis accompanied by enthesitis into juvenile PsA and ERA, particular care will be required to avoid classification of patients with juvenile PsA into ERA. Current ILAR criteria attempt this division by excluding from ERA patients with a personal or first-degree family history of psoriasis. Our data suggest that still more stringent qualifications are necessary, such as the exclusion of patients with nail pits or dactylitis.

Our report is based on children diagnosed with juvenile PsA at a single center and reflects the diagnostic practices of one group of pediatric rheumatologists. Despite this limitation, our findings raise doubt as to the ability of current ILAR criteria to accurately classify children with PsA. It is not our intention to advocate a return to the use of the Vancouver criteria: the term psoriasis-like rash is vague, a second-degree family history of psoriasis is of limited specificity and questionable relevance, and the relegation of nail pits and dactylitis to minor criteria likely understates their diagnostic value among children with arthritis. Instead, we suggest that investigators maintain careful phenotypic data on their patients and consider classifying patients by both Vancouver and ILAR criteria until relevant pathophysiologic subclasses within these populations are better understood and improved classification criteria are developed.

In conclusion, we find that strict application of the ILAR criteria to a population of patients with Vancouver-defined juvenile PsA results in the inappropriate partition of the majority of these patients to other categories within the JIA nomenclature. We believe that this result is of substantial importance to the field of pediatric rheumatology. Until the juvenile arthritides may be subdivided on the basis of pathophysiology, careful phenotypic classification is critical. Our findings strongly suggest that the ILAR criteria do not properly demarcate the population of children with juvenile PsA, and as a corollary risk cross-contaminating other diagnostic categories within the JIA system with patients better regarded as having juvenile PsA. As a result, the criteria as they stand are likely to complicate the investigation of juvenile PsA as well as the JIA categories into which patients with juvenile PsA are inappropriately included.


Dr. Nigrovic had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Stoll, Sundel, Nigrovic.

Acquisition of data. Stoll, Lio, Nigrovic.

Analysis and interpretation of data. Stoll, Lio, Nigrovic.

Manuscript preparation. Stoll, Lio, Sundel, Nigrovic.

Statistical analysis. Stoll.


We are indebted to Robert C. Fuhlbrigge, MD, PhD, and Bryce Binstadt, MD, PhD, for thoughtful comments on the manuscript.